Method / Two layers

Process / Reading framework

Turn a feeling that something is off into something you can judge, train and deliver.

This page has two layers: a five-step process for how collaboration runs, and a six-step framework for how a reading is done.

Fill in the screening form See the research basis →

Process

Five steps from intake to written findings

Intake and scope

A written intake captures the material type, language, use context and the decision the review should support. Scope, deliverable format and timeline are confirmed in writing before close reading begins.

Close reading and annotation

Anonymized material is read in full. Pattern markers cover authority flow, role position, semantic drift points, framing choices, uncertainty handling and affective rhythm.

Pattern mapping

Annotated observations are organized into recurring patterns, drift sequences and points where user interpretation may diverge from system intent.

Findings memo

A written memorandum presents observations, risk localization, framing recommendations and evidence level notes. Optional artifacts include annotated transcripts and risk matrices.

Clarification

A short asynchronous round addresses follow up questions. Further work is scoped separately when additional review is requested.

Source material is deleted within 30 days of delivery. One in-scope revision is included at no extra charge.

Framework

The six-step reading

1 Signal

Question: how dangerous is this conversation? Output: a risk level (low, medium, high, critical) and a signal list.

2 Context

Question: which conversation is actually happening? Output: a reading of the surface issue versus the real issue.

3 Consequence

Question: where does each option lead? Output: candidate responses mapped to consequence paths.

4 Strategy

Question: speak, stay silent, or change the channel? Output: a chosen strategy with a reason you can say out loud.

5 Wording

Question: what exactly to say, in what order? Output: an executable script and sequence, with a fallback version.

6 Calibration

Question: after the conversation, did the criteria get sharper? Output: a one-page update to a personal reading handbook.

The same framework applies to human and AI dialogue: the signal layer adds AI accommodation and overconfidence, and the context layer adds “the AI has no real position, but you projected one”.

Evidence discipline

Four evidence roles, labeled as they are

Inclusion, gray-zone, negative and protective gray-zone cases are labeled as what they are. Gray zones are not forced into a verdict.

Findings are tied to evidence and stated with the appropriate level of confidence. Prevalence estimates, clinical validity claims, psychometric validation and intervention models sit outside the current evidence level.

Outcomes

What each step buys you

Signal

Early warning before the incident.

Context

Fewer misreadings, less internal friction.

Consequence

Decisions with evidence that can be reviewed.

Strategy

Criteria for hard conversations.

Wording

Stable quality when you act.

Calibration

Capability stays with the organization.

Sources

Where the method comes from

The six-step reading is the consulting delivery form of three frameworks: CXC-7, USCP and USCI.

CXC-7

https://doi.org/10.5281/zenodo.17247637

USCP

https://doi.org/10.5281/zenodo.21346372

USCI

https://doi.org/10.5281/zenodo.18678196

Materials

What can be reviewed

Dialogue

Real or simulated dialogue transcripts. Chatbot flows. Prompt sets.

Output

Cross language AI output. Response test sets.

Documents

Course copy and AI literacy materials. Research summaries and public facing reports.

Begin

Describe your material and the decision it should support, in the screening form.

Fill in the screening form